Mental health spillovers in primary schools

In this paper we use extensive Danish register data to study peer effects from students with mental health problems. In our analysis, mental health problems are defined from health information in registers on medical prescriptions, contact with general practitioners and psychologist and psychiatrists. We follow classes from 1st to 6th grade between 2007 and 2013 and study mental health trajectories of incumbent students when receiving a new student with mental health problems. We use a time to event design, controlling for individual and grade fixed effects and using never treated classes as the control group. We condition on students being observed from grade 1 until treatment and follow them after treatment irrespectively of the school or class they attend. Following the new difference-in-differences literature, we produce estimates that are robust to heterogeneity across cohorts and grades. As additional robustness check, we compare against classes that receive a new student without a mental health problem, and produce results adjusting for changes in diagnoses trends. We find that outcomes for incumbent students are affected by receiving a new student with mental health problems, an effect that is larger when first exposure occurs by grade 6. These effects are long-lasting irrespective of the gender of the mover and of the gender of the incumbent student. We are currently working with well-being data to identify whether the effect works through direct student-to-student contagion (behavioral effect) or through a higher likelihood of diagnosis due to parental awareness of the condition (information effect).